src/RecfileCommand.h
author František Kučera <franta-hg@frantovo.cz>
Sun, 07 Apr 2019 01:10:31 +0200
branchv_0
changeset 3 891fe13d7397
parent 2 2390e2949a36
child 4 b5239b4b345b
permissions -rw-r--r--
handler logic structure
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     1
/**
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     2
 * Relational pipes
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     3
 * Copyright © 2019 František Kučera (Frantovo.cz, GlobalCode.info)
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     4
 *
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     5
 * This program is free software: you can redistribute it and/or modify
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     6
 * it under the terms of the GNU General Public License as published by
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     7
 * the Free Software Foundation, either version 3 of the License, or
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     8
 * (at your option) any later version.
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
     9
 *
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    10
 * This program is distributed in the hope that it will be useful,
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    11
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    12
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    13
 * GNU General Public License for more details.
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    14
 *
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    15
 * You should have received a copy of the GNU General Public License
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    16
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    17
 */
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    18
#pragma once
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    19
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    20
#include <iostream>
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    21
#include <string>
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    22
#include <sstream>
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    23
#include <vector>
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    24
#include <set>
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    25
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    26
#include <relpipe/writer/typedefs.h>
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    27
#include <relpipe/writer/RelationalWriter.h>
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    28
#include <relpipe/writer/AttributeMetadata.h>
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    29
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    30
namespace relpipe {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    31
namespace in {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    32
namespace recfile {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    33
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    34
using namespace relpipe::writer;
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    35
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    36
class RecfileCommand {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    37
private:
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    38
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    39
	enum class RecfileLineType {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    40
		METADATA,
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
    41
		DATA,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
    42
		SEPARATOR,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
    43
		COMMENT,
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    44
		END,
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    45
	};
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    46
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    47
	class RecfileHandler {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    48
	private:
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
    49
		RelationalWriter* writer;
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    50
		string_t currentRelationName;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    51
		std::vector<string_t> currentRecord;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    52
		std::vector<std::vector<string_t>> currentRecords;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    53
		size_t prefetchCount = 1;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    54
		bool headerWritten = false;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    55
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    56
		void writeHeader() {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    57
			if (headerWritten) return;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    58
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    59
			std::vector<AttributeMetadata> attributeMetadata;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    60
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    61
			// TODO: writer->startRelation(currentRelationName, attributeMetadata, true);
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    62
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    63
			headerWritten = true;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    64
		}
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    65
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    66
		void writeRecords() {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    67
			for (int i = 0; i < currentRecords.size(); i++) {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    68
				std::vector<string_t> record = currentRecords[i];
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    69
				std::wcerr << L"record: " << std::endl;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    70
				for (int j = 0; j < record.size(); j += 2) {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    71
					std::wcerr << L"  '" << record[j] << L"': '" << record[j + 1] << L"'" << std::endl;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    72
					// TODO: writer->writeAttribute()
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    73
				}
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    74
			}
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    75
			currentRecords.clear();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    76
		}
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    77
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    78
		void metadata(const string_t& name, const string_t& value) {
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    79
			std::wcerr << L"metadata(" << name << L", " << value << L");" << std::endl; // TODO: remove debug
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    80
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    81
			if (name == L"rec") {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    82
				currentRelationName = value;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    83
				currentRecord.clear();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    84
				currentRecords.clear();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    85
				headerWritten = false;
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    86
			} else if (name == L"type") {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    87
				// TODO: save type hint
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    88
			} else {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    89
				// ignore – other recfile metadata like keys or auto-increments
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    90
			}
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    91
		}
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    92
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    93
		void data(const string_t& name, const string_t& value) {
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    94
			std::wcerr << L"data(" << name << L", " << value << L");" << std::endl; // TODO: remove debug
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    95
			currentRecord.push_back(name);
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
    96
			currentRecord.push_back(value);
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    97
		}
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    98
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
    99
		void comment(const string_t& value) {
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   100
			// ignore comments
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   101
		}
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   102
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   103
		void separator() {
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   104
			std::wcerr << L"separator()" << std::endl; // TODO: remove debug
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   105
			if (currentRecord.size()) {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   106
				currentRecords.push_back(currentRecord);
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   107
				currentRecord.clear();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   108
			}
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   109
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   110
			if (prefetchCount > 0 && currentRecords.size() >= prefetchCount) {
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   111
				writeHeader();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   112
				writeRecords();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   113
			}
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   114
		}
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   115
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   116
		void end() {
3
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   117
			std::wcerr << L"end();" << std::endl; // TODO: remove debug
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   118
			if (currentRecord.size()) currentRecords.push_back(currentRecord);
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   119
			writeHeader();
891fe13d7397 handler logic structure
František Kučera <franta-hg@frantovo.cz>
parents: 2
diff changeset
   120
			writeRecords();
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   121
		}
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   122
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   123
	public:
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   124
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   125
		RecfileHandler(RelationalWriter* writer) : writer(writer) {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   126
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   127
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   128
		virtual ~RecfileHandler() {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   129
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   130
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   131
		void logicalLine(RecfileLineType type, const string_t& name = L"", const string_t& value = L"") {
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   132
			switch (type) {
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   133
				case RecfileLineType::METADATA: return metadata(name, value);
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   134
				case RecfileLineType::DATA: return data(name, value);
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   135
				case RecfileLineType::COMMENT: return comment(value);
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   136
				case RecfileLineType::SEPARATOR: return separator();
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   137
				case RecfileLineType::END: return end();
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   138
			}
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   139
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   140
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   141
	};
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   142
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   143
	enum class ParserState {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   144
		START,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   145
		NAME,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   146
		VALUE,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   147
		VALUE_CONTINUATION,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   148
		COMMENT,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   149
		END,
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   150
	};
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   151
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   152
	class RecfileParser {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   153
	private:
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   154
		wstring_convert<codecvt_utf8<wchar_t>> convertor; // TODO: support also other encodings or are recfiles always in UTF-8?
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   155
		RecfileHandler& handler;
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   156
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   157
		void emitLogicalLine(RecfileLineType& type, std::stringstream& name, std::stringstream& value) {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   158
			handler.logicalLine(type, convertor.from_bytes(name.str()), convertor.from_bytes(value.str()));
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   159
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   160
			name.str("");
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   161
			name.clear();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   162
			value.str("");
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   163
			value.clear();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   164
			type = RecfileLineType::DATA;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   165
		}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   166
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   167
	public:
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   168
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   169
		RecfileParser(RecfileHandler& handler) : handler(handler) {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   170
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   171
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   172
		virtual ~RecfileParser() {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   173
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   174
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   175
		void parse(std::istream& input) {
1
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   176
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   177
			ParserState state = ParserState::START;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   178
			RecfileLineType type = RecfileLineType::DATA;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   179
			std::stringstream name;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   180
			std::stringstream value;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   181
			char ch;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   182
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   183
			while (state != ParserState::END && input.good()) {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   184
				ch = input.get();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   185
				if (input.eof()) continue;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   186
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   187
				switch (state) {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   188
					case ParserState::START:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   189
						if (ch == '%') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   190
							type = RecfileLineType::METADATA;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   191
							break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   192
						} else if (ch == ' ') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   193
							break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   194
						} else if (ch == '\n') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   195
							handler.logicalLine(RecfileLineType::SEPARATOR);
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   196
							break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   197
						} else if (ch == '#') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   198
							type = RecfileLineType::COMMENT;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   199
							state = ParserState::COMMENT;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   200
							if (input.get() != ' ') input.unget();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   201
							break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   202
						} // else → name
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   203
					case ParserState::NAME:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   204
						if (ch == ':') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   205
							state = ParserState::VALUE;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   206
							if (input.get() != ' ') input.unget();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   207
						} else {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   208
							name << ch;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   209
						}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   210
						break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   211
					case ParserState::VALUE:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   212
						if (ch == '\n') state = ParserState::VALUE_CONTINUATION;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   213
						else value << ch;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   214
						break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   215
					case ParserState::VALUE_CONTINUATION:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   216
						if (ch == '+') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   217
							state = ParserState::VALUE;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   218
							if (value.tellp()) value << '\n';
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   219
							if (input.get() != ' ') input.unget();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   220
						} else {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   221
							input.unget();
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   222
							state = ParserState::START;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   223
							emitLogicalLine(type, name, value);
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   224
						}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   225
						break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   226
					case ParserState::COMMENT:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   227
						if (ch == '\n') {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   228
							state = ParserState::START;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   229
							emitLogicalLine(type, name, value);
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   230
						} else {
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   231
							value << ch;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   232
						}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   233
						break;
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   234
					default:
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   235
						throw RelpipeWriterException(L"Unknown ParserState: " + std::to_wstring((int) state) + L" in RecfileParser."); // TODO: better exception
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   236
				}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   237
			}
8dfb42e5c088 parse recfile (logical lines)
František Kučera <franta-hg@frantovo.cz>
parents: 0
diff changeset
   238
			emitLogicalLine(type, name, value);
2
2390e2949a36 parse recfile: separate methods
František Kučera <franta-hg@frantovo.cz>
parents: 1
diff changeset
   239
			handler.logicalLine(RecfileLineType::END);
0
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   240
		}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   241
	};
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   242
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   243
public:
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   244
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   245
	void process(std::istream& input, std::ostream& output) {
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   246
		unique_ptr<RelationalWriter> writer(Factory::create(output));
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   247
		RecfileHandler handler(writer.get());
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   248
		RecfileParser parser(handler);
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   249
		parser.parse(input);
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   250
	}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   251
};
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   252
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   253
}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   254
}
515a697cc9cd project skeleton
František Kučera <franta-hg@frantovo.cz>
parents:
diff changeset
   255
}